Stefan Bauer
About
Current AI systems are limited in their ability to understand the world around us, as shown in a limited ability to transfer to new problems or lack of skill in applying known tools in an unknown scenario.
This is a shared problem for many established approaches, which mostly focus on predictability when inferring patterns and structure from data. Stefan Bauer’s key research goal is to design machines that can extrapolate experience across environments and tasks by learning independent mechanisms that can flexibly be used, composed and re-purposed.
Awards
- 2019 Best Paper, International Conference for Machine Learning (ICML)
- 2018 ETH Medal for Outstanding Doctoral Thesis
Relevant Publications
Pfister, N., Bauer, S., & Peters, J. (2019). Learning stable and predictive structures in kinetic systems. Proceedings of the National Academy of Sciences, 116(51), 25405-25411. doi:10.1073/pnas.1905688116
Suter, R., Miladinovic, D., Schölkopf, B., & Bauer, S. (2019, May). Robustly disentangled causal mechanisms: Validating deep representations for interventional robustness. In International Conference on Machine Learning (pp. 6056-6065).
Locatello, F., Bauer, S., Lucic, M., Rätsch, G., Gelly, S., Schölkopf, B., & Bachem, O. (2019, May). Challenging common assumptions in the unsupervised learning of disentangled representations. In International Conference on Machine Learning (pp. 4114-4124).